# Vidmoat models

> The models Vidmoat runs on its own infrastructure, who can use them, what they cost, and how to call one in three requests.

Most of Vidmoat's generation is routed to outside providers. These models are different: we run them ourselves, on our own servers and GPUs, behind our own checks. That means we control the price, the limits and how each one behaves, and each one ships with a model card saying what it does and what it will not do.

| Model | What it does | Price | Status |
| --- | --- | --- | --- |
| [`vidmoat-image-live-1`](https://developer.vidmoat.com/developer/docs/models/vidmoat-image-live-1) | Turns a still into a short living clip: camera moves, moving water, smoke and leaves, people breathing and moving. Nothing new is added to the picture. | 8 or 24 credits a second | Generally available |

> **Tip: Browse every model.** The [Models catalogue](https://developer.vidmoat.com/developer/models) lists every model you can call through Vidmoat, ours and our partners’, with search and filters by category, maker and surface.

- [vidmoat-image-live-1](https://developer.vidmoat.com/developer/docs/models/vidmoat-image-live-1): The model card: tiers, inputs, prices, speed, safety checks and known limits.
- [Bring a still to life](https://developer.vidmoat.com/developer/docs/api/generation/create-live-image): The API reference for `POST /v1/ai/live-images`, with every parameter and error.

## Who can use them

| Where | Who | How |
| --- | --- | --- |
| The API | Live keys: Studio, Team and Enterprise. Test keys: any plan, free, returning a sample job. | `POST /v1/ai/live-images` with a key that has the `ai.video` scope. |
| MCP (Claude, ChatGPT, Cursor and others) | Creator and every plan above it. | The `generate_live_image` tool. |
| The Vidmoat editor agent | Creator and every plan above it. | Ask it to bring a still to life. |

Every call is billed in credits to the account that owns the key or the sign-in, at the same price everywhere. Hobby accounts get `402 plan_required` and are charged nothing.

### Step 1: Create an app and a key

In the [developer console](https://developer.vidmoat.com/developer/apps), create an app and a key with the `ai.video` scope. Start with a test key: it costs nothing and returns a sample job, so you can build the whole flow first.

### Step 2: Ask for a quote

```bash
curl -X POST https://api.vidmoat.com/v1/ai/live-images \
  -H "Authorization: Bearer $VIDMOAT_KEY" \
  -H "Content-Type: application/json" \
  -d '{"image":{"url":"https://example.com/lake.jpg"},"prompt":"slow push-in, the water ripples","quoteOnly":true}'
```

The quote names the tier, the method and the exact credits. Nothing is charged, and the quote is the charge.

### Step 3: Start the job and poll it

Send the same body without `quoteOnly`. You get `202` with a job `id`; poll [`GET /v1/ai/jobs/{id}`](https://developer.vidmoat.com/developer/docs/api/generation/get-generation-job) every five seconds or so until `status` is `COMPLETED`, then download `outputUrl`. A job that fails is refunded in full.

> **Note: Versions.** A model name never changes meaning. When a model is retrained in a way you would notice, it ships under a new name (`vidmoat-image-live-2`) and the old one keeps working until we announce a retirement date in the [changelog](https://developer.vidmoat.com/developer/docs/changelog).

---

Source: https://developer.vidmoat.com/developer/docs/models
Previous: [Update a bot user](https://developer.vidmoat.com/developer/docs/api/telegram-users/update-telegram-user.md)
Next: [vidmoat-image-live-1](https://developer.vidmoat.com/developer/docs/models/vidmoat-image-live-1.md)
All documentation: https://developer.vidmoat.com/llms-full.txt